Supplementary Figure 3: Biostatistical modeling of the clonal fate data.

(a) The likelihood function F gives the probability of the experimental data for different values of the induction frequency pN and the fragmentation rate f. The numeric values have been rescaled such that the maximum of the likelihood function corresponds to 1. Colour denotes the value of F, such that red signifies a large value and blue signifies small value. Lines of equal values are indicated on the bottom of the figure. One sees that the maximum value of F is relatively featureless along a curve in the pN-f-plane. To infer the values of pN and f we must therefore refer to an independent measurement of one of the two parameters. (b) The multicolour labelling strategy allows us to independently infer the induction frequency pN = 1.3 by evaluating the abundances of hearts with a given number of colours. With this, we are left with a slice through the pN-f-plane and the fragmentation rate can be determined with a higher accuracy. (c) Monoclonal datasets (n = 89) identify two subpopulations in Mesp1 expressing cells: FHF progenitors, which contribute to the LV and SHF progenitors, which contribute to OFT and IFT. The plot shows the probabilities of monoclonal fragments in the different heart compartments. (d) Values for the induction frequency, pN, and the fragmentation rate, f, for the two FHF (n = 188) and SHF (n = 102) precursors. While the overall induction frequency is higher for FHF precursors, which we attribute to highest expression of Mesp1 at the early time points, the fragmentation rate is higher for SHF precursors. (e) We may use the distribution of monoclonal fragments (c) to predict the distribution of fragments in all hearts (n = 263). We find an excellent agreement with the notable exception of the RV, which might suggest the existence of an independent pool of progenitors contributing to RV morphogenesis. Error bars indicate one sigma (c and e) or 95% (d) confidence intervals.